[plt-scheme] structure inspectors

From: Nicholas Chubrich (chubrich at cs.brandeis.edu)
Date: Mon May 22 11:58:06 EDT 2006

How do you go about swapping the inspector?  I have this code:
(current-inspector (make-inspector))
(require (lib "mred.ss" "mred"))
(require (lib "class.ss"))
(define f (instantiate frame% ("Simple Edit" #f 200 200)))
but if I then call for instance class-info on frame% I still get an error 
message that my current inspector cannot inspect.
	Come to think of it, is there any way to reference another 
inspector other than make-inspector?

On Mon, 22 May 2006, Matthew Flatt wrote:

> At Mon, 22 May 2006 08:11:01 -0400 (EDT), Nicholas Chubrich wrote:
> > I would like to examine a built-in structure, specifically the-style-list 
> > in the MrEd editor toolkit.  This means dealing with inspectors.  I saw in 
> > the list archive that you could use a procedure called make-->vector to 
> > bypass inspectors, but I cannot get this to work, because I am unable to 
> > come up with a struct type descriptor.  It should be struct:style-list%, 
> > but this ends up being an unbound value in the environment.
> > 	Any suggestions?  I would like to be able to look at style lists 
> > in a single glance, without having to access each field individually.
> A style list is an object in the sense of `(lib "class.ss")'. So, it's
> a structure, but it doesn't correspond to a simple `define-struct'.
> You could get a sufficiently powerful inspector by swapping the
> inspector before `(lib "mred.ss" "mred")' is loaded, which means using
> the -Z command-line argument. But this is surely not the direction you
> want to go. For one thing, the object mostly just contains a pointer to
> opaque "primitive" data, combined with the methods to extract parts of
> the data. Even if the native data was inspectable, the implementation
> of a style list is too complicated for the raw object content to be
> informative.
> In short, I think the only useful thing for a `style-list%' object is
> an explicit print function that uses the existing methods.
> Matthew

Posted on the users mailing list.